Deuterium Isotope Effects on Hydrophobic Interactions: The

Dalit Rechavi, Alessandro Scarso, and Julius Rebek, Jr. .... Leanne B. Ohlund , Tze Chieh Shiao , Amélie Vézina , Borhane Annabi , René Roy , Lekha...
5 downloads 0 Views 212KB Size
Download PDF
Published on Web 10/17/2003

Deuterium Isotope Effects on Hydrophobic Interactions: The Importance of Dispersion Interactions in the Hydrophobic Phase Maciej Turowski,†,‡ Naoki Yamakawa,† Jaroslaw Meller,§ Kazuhiro Kimata,† Tohru Ikegami,† Ken Hosoya,† Nobuo Tanaka,*,† and Edward R. Thornton*,⊥ Contribution from the Department of Polymer Science and Engineering, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan, and the Department of Chemistry, UniVersity of PennsylVania, Philadelphia, PennsylVania 19104-6323 Received May 7, 2003; Revised Manuscript Received July 28, 2003; E-mail: [email protected]; [email protected].

Abstract: Hydrogen/deuterium isotope effects on hydrophobic binding were examined by means of reversedphase chromatographic separation of protiated and deuterated isotopologue pairs for a set of 10 nonpolar and low-polarity compounds with 10 stationary phases having alkyl and aryl groups bonded to the silica surface. It was found that protiated compounds bind to nonpolar moieties attached to silica more strongly than deuterated ones, demonstrating that the CH/CD bonds of the solutes are weakened or have less restricted motions when bound in the stationary phase compared with the aqueous solvent (mobile phase). The interactions responsible for binding have been further characterized by studies of the effects of changes in mobile phase composition, temperature dependence of binding, and QSRR (quantitative structurechromatographic retention relationship) analysis, demonstrating the importance of enthalpic effects in binding and differentiation between the isotopologues. To explain our results showing the active role of the hydrophobic (stationary) phase we propose a plausible model that includes specific contributions from aromatic edge-to-face attractive interactions and attractive interactions of aliphatic groups with the π clouds of aromatic groups present as the solute or in the stationary phase.

Introduction

Reversed-phase, high-performance liquid chromatography (RPLC) involves transfer of solute from a polar, aqueous mobile phase to a nonpolar, hydrophobic stationary phase. Partitioning between mobile and stationary phase has been shown to be the predominant retention mechanism, and thus RPLC constitutes a useful model of hydrophobic effects that are of great importance in biology and chemistry. The opportunity exists to capitalize on the very high precision of RPLC to measure isotope effects, which constitute a unique probe of molecular interactions, upon transfer of hydrophobic molecules from an aqueous to a nonpolar phase. We have previously shown the utility of RPLC for highly precise isotope effect studies.1-4 Advances in understanding of * Corresponding authors. † Kyoto Institute of Technology. ‡ On leave from the Department of Biopharmaceutics and Pharmacodynamics, Medical University of Gdansk, ul. Gen. J. Hallera 107, PL80-416 Gdansk, Poland. Present address: The Dow Chemical Company, Analytical Sciences, Midland, MI 48667. E-mail: [email protected]. § Department of Synthetic and Biological Chemistry, Kyoto University, Kyoto, Japan. Present address: Children’s Hospital Research Foundation, Pediatric Informatics, Cincinnati, OH 45229. E-mail: [email protected]. ⊥ University of Pennsylvania. (1) (a) Tanaka, N.; Thornton, E. J. Am. Chem. Soc. 1976, 98, 1617-1619. (b) Tanaka, N.; Thornton, E. J. Am. Chem. Soc. 1977, 99, 7300-7306. (c) Lockley, W. J. S. J. Chromatogr. 1989, 483, 413-418. (2) Tanaka, N.; Hosoya, K.; Nomura, K.; Yoshimura, T.; Ohki, T.; Yamaoka, R.; Kimata, K.; Araki, M. Nature 1989, 341, 727-728. (3) Kimata, K.; Hosoya, K.; Araki, T.; Tanaka, N. Anal. Chem. 1997, 69, 26102612. 13836

9

J. AM. CHEM. SOC. 2003, 125, 13836-13849

hydrophobic effects in general and, simultaneously, of the RPLC separation process should be accessible through further study of RPLC isotope effects over a wider range of stationary phase structures, solute structures, and mobile-phase compositions. Hydrophobic interactions play a key role in determining the structure and function of lipid membranes5 and proteins6 as well as the activity of the drugs and toxins.7,8 Several techniques for (4) See also: Cartoni, G. P.; Ferreti, I. J. Chromatogr. 1976, 122, 287. (5) (a) Ne´methy, G.; Scheraga, H. A. J. Chem. Phys. 1962, 36, 3382-3400, 3401-3417. (b) Scheraga, H. A. J. Biomol. Struct. Dyn. 1998, 16, 447460. (c) Tanford, C. Science 1978, 200, 1012-1018. (d) Tanford, C. The Hydrophobic Effect: Formation of Micelles and Biological Membranes, 2nd ed.; Wiley: New York, 1980; and references therein. (e) Kyte, J. Biophys. Chem. 2003, 100, 193-202. (f) Southall, N. T.; Dill, K. A.; Haymet, A. D. J. J. Phys. Chem. B 2002, 106, 521-533. (g) Southall, N. T.; Dill, K. A. J. Phys. Chem. B 2000, 104, 1326-1331. (h) Huang, D. M.; Chandler, D. J. Phys. Chem. B 2002, 106, 2047-2053. (i) Hummer, G.; Garde, S.; Garcı´a, A. E.; Pratt, L. R. Chem. Phys. 2000, 258, 349370. (j) Soda, K. AdV. Biophys. 1993, 29, 1-54. (k) Waldbillig, R. C.; Szabo, G. Biochim. Biophys. Acta 1979, 557, 295-305. (l) Helm, C. A.; Israelachvili, J. N.; McGuiggan, P. M. Science 1989, 246, 919-922. (m) Helm, C. A.; Israelachvili, J. N.; McGuiggan, P. M. Biochemistry 1992, 31, 1794-1805. (6) (a) Branden, C.; Tooze, J. Introduction to Protein Structure, 2nd ed.; Garland Publ.: New York, 1998. (b) Ne´methy, G.; Scheraga, H. A. J. Phys. Chem. 1962, 66, 1773-1789; Erratum: J. Phys. Chem. 1963, 67, 2888. (c) Takano, K.; Yamagata, Y.; Yutani, K. J. Mol. Biol. 1998, 280 (4), 749761. (d) Baldwin, R. L. Proc. Natl. Acad. Sci. U.S.A. 1986, 83 (21), 80698072. (e) To´th, G.; Murphy, R. F.; Lovas, S. Protein Eng. 2001, 14, 543547. (7) Balaz, S. Membrane Transport and Cellular Distribution. In: Lipophilicity in Drug Design and Toxicology; Methods and Principles in Medicinal Chemistry; Pliska, V., Testa, B., van de Waterbeemd, H., Eds.; VCH: Weinheim, 1996; Vol. 4, pp 295-308. (8) Hansch, C.; Kim, D.; Leo, A. J.; Novellino, E.; Silipo, C.; Vittoria, A. Crit. ReV. Toxicol. 1989, 19, 185-226. 10.1021/ja036006g CCC: $25.00 © 2003 American Chemical Society

Isotope Effects on Hydrophobic Interactions

evaluation of hydrophobic properties are available, including measurement of n-octanol/water partition coefficients P for a series of compounds.9 RPLC10,11 methods have provided an excellent method to give reliable hydrophobicity measurements.12 H/D isotope effects are widely used to characterize chemical processes.13-16 RPLC separations of protiated and deuterated pairs of compounds1,4 and separation of nitrogen and oxygen isotopes in acids and bases2 have been reported. Tritium isotope effects in high-performance liquid chromatography (HPLC) of eicosanoids and vitamin D metabolites have also been reported.17,18 Recently, separation of enantiomers based on isotopic chirality was described.3 A cellulose-based stationary phase was shown to separate racemic phenyl(phenyl-d5)methanol, thus demonstrating that the chiral stationary phase exhibits different interactions with the two enantiomers even though they differ only by the stereochemical positioning of phenyl and phenyld5. These results show that liquid chromatography techniques constitute highly convenient and sensitive tools for precise study of isotope effects. In this work we have extensively examined secondary isotope effects in the retention process of RPLC, as a means of investigating the nature of hydrophobic effects. In addition, both HPLC and mass spectrometric (MS) methods such as isotopecoded affinity are used to compare protein profiles of different cells, e.g., diseased vs healthy.19,20 The ability to separate isotopes by HPLC suggests the possibility of combining current LC-MS methods with isotope-coded differential LC to provide a new method for measurement of differential protein content. The nature of isotope effects involves to a large extent vibrational frequencies.21 Within the commonly accepted BornOppenheimer approximation, the electronic wave functions and the resulting potential energy surfaces do not change upon isotopic substitution but the vibrational wave functions of deuterium and protium are different because the vibrational states depend on the mass of the nuclei. Ordinarily, structural changes in solute, solvent, or stationary phase will involve changes in potential energies of interaction which are very (9) (a) Hansch, C.; Maloney, P. P.; Fujita, T.; Muir, R. M. Nature 1962, 194, 178-180. (b) Fujita, T.; Iwasa, J.; Hansch, C. J. Am. Chem. Soc. 1964, 86, 5175-5180. (c) Hansch, C.; Leo, A.; Hoekman, D. Hydrophobic, Electronic, and Steric Constants; ACS Professional Reference Books: Washington, DC, 1995; Exploring QSAR, Vol. 2. (10) (a) Braumann, T. J. Chromatogr. 1986, 373, 191-225. (b) Dorsey, J. G.; Khaledi, M. G. J. Chromatogr. 1993, 656, 485-499. (11) (a) Kaliszan, R. QuantitatiVe Structure-Chromatographic Retention Relationships; John Wiley & Sons: New York, 1987. (b) Nasal, A.; Sznitowska, M.; Bucinski, A.; Kaliszan, R. J. Chromatogr. A 1995, 692, 83-89. (12) (a) Krass, J. D.; Jastorff, B.; Genieser, H. G. Anal. Chem. 1997, 69, 25752581. (b) Valko´, K.; Bevan, C.; Reynolds, D. Anal. Chem. 1997, 69, 20222029. (13) Streitwieser, A., Jr.; Klein, H. S. J. Am. Chem. Soc. 1964, 86, 5170-5173. (14) Wolfsberg, M. Acc. Chem. Res. 1972, 5, 225-233. (15) Isotope Effects in Chemical Reactions, ACS Monograph; Collins, C. J., Bowman, N. S., Eds.; Van Nostrand Reinhold Company: New York, 1970. (16) Streitwieser, A., Jr.; Klein, H. S. J. Am. Chem. Soc. 1963, 85, 2759-2763. (17) Do, U. H.; Lo, S. L.; Iles, J.; Rosenberger, T.; Tam, P.; Hong, Y.; Ahern, D. Prostagl. Leucotri. Ess. Fatty Acids 1994, 50, 335-338. (18) Worth, G. K.; Retallack, R. W. Anal. Biochem. 1988, 174, 137-141. (19) Partial separation of istopically labeled peptides has been demonstrated; it was shown that such incomplete separation could cause significant systematic errors in quantitative proteomics: Zhang, R.; Sioma, C. S.; Wang, S.; Regnier, F. E. Anal. Chem. 2001, 73, 5142-5149. (20) Yu, L. R.; Johnson, M. D.; Conrads, T. P.; Smith, R. D.; Morrison, R. S.; Veenstra, T. D. Electrophoresis 2002, 23, 1591-1598. (21) (a) Thornton, E. R. Annu. ReV. Phys. Chem. 1966, 17, 349-372. (b) Bartell, L. S. Tetrahedron Lett. 1960, 6, 13-16. (c) Halevi, E. A.; Nussim, M.; Ron, A. J. Chem. Soc. 1963, 866-875.

ARTICLES

difficult to estimate theoretically, so that interpretation of binding effects is also very difficult. But since isotopic substitution does not affect potential energy, potential energy effects cancel between isotopologues, and isotope effects involve only the effects of changes in interactions upon nuclear motions, especially upon vibrational frequencies. Consequently, isotope effects probe changes in molecular interactions without involving the difficulties of interpreting potential energy changes. Isotope effects therefore provide a unique and readily interpreted experimental means of investigating molecular interactions. As a result of its larger mass, the amplitude of vibrations is smaller for deuterium, also resulting in slightly lower average volumes and polarizabilities for bonds involving deuterium than for the corresponding bonds involving protium.22 We use calculations of these effects to give additional insights into the isotope effects we have observed. Measuring retention factors in RPLC gives direct information about the free energy associated with transfer of a nonpolar solute from an aqueous mobile phase to an organic stationary phase, corresponding to the binding process, and the effect of isotopic substitution on this process (cf. eq 1-5 below). Though a complete description of the hydrophobic binding process is still under development,5 fundamental work has related hydrophobic effects to retention mechanisms in liquid chromatography.23-27 RPLC involves an equilibrated aqueous mobile phase and an organic stationary phase, the partitioning process being controlled by both mobile-phase and stationaryphase effects. Mutual van der Waals interactions between the nonpolar stationary phase and solutes operate, in particular, dispersion forces, resulting in attractive binding of solute within the stationary phase. Because of the unique ability of isotope effects to probe the nature of solute interactions, our studies allow us to dissect the total hydrophobic process into contributions from mobile phase hydrophobic phenomena and from binding properties of the stationary phase. Simulation techniques, such as the free energy perturbation method with commonly used molecular dynamics force fields,28 are usually unable to reproduce the experimental numbers with satisfactory accuracy for the purposes of analyzing HPLC retention processes. Because such difficulties are inherent in accurately estimating small changes in interaction energies, we employ here quantitative structure-chromatographic retention relationships11 (QSRR) analysis, an extrathermodynamic linear free energy relationship (LFER), to aid in elucidating the nature of hydrophobic binding in RPLC.29 The stationary phases tend to resemble organic liquid phases but differ in that they possess significant ordering resulting from the attachment of the hydrophobic chains to the silica particle core.25 These stationary phases will bind the methanol component of the mobile phase to some extent, so the properties we measure refer to a methanol-saturated stationary phase. Almost (22) Deuterium isotope effects in noncovalent interactions have been reviewed: Wade, D. Chem.-Biol. Interact. 1999, 117, 191-217. (23) Tan, L. C.; Carr, P. W. J. Chromatogr. A 1997, 775, 1-12. (24) Vailaya, A.; Horva´th, Cs. J. Chromatogr. A 1998, 829, 1-27. (25) (a) Dorsey, J. G.; Dill, K. A. Chem. ReV. 1989, 89, 331-346. (b) Dorsey, J. G.; Cooper, W. T. Anal. Chem. 1994, 66 (17), 857A-867A. (26) Jaroniec, M. J. Chromatogr. 1993, 656, 37-50. (27) Zhao, J.; Carr, P. W. Anal. Chem. 1998, 70, 3619-3628. (28) Meller, J. Molecular Dynamics. http://folding.chmcc.org/publications/ 3048.doc. In Encyclopedia of Life Sciences; http://www.els.net. Macmillan Reference Ltd., 2001. (29) Klatte, J. S.; Beck, T. L. J. Phys. Chem. 1996, 100, 5931-5934. J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003 13837

ARTICLES

Turowski et al. Materials and Methods. Deuterated compounds were available from Aldrich or CEA (Commisariat a` L’Energie Atomique, France). 1-Decand21-ol was prepared from decanoic-d19 acid, 1-pentan-d11-ol was prepared from pentanoic-d9 acid, and phenyl-d5-methyl alcohol was prepared from the benzoic-d5 acid. All other chemicals were of analytical grade and were available from the common major suppliers. We employed a variety of structures for the stationary phase materials (Figure 1), all based on silica bonded beads: aliphatic octyl C8 and octadecyl C18, fluorinated 4,4-di(trifluoromethyl)-5,5,6,6,7,7,7-heptafluoroheptyl, F13C9, highly dispersive 3-(pentabromobenzyloxy)propyl, PBB, highly aromatic 2-(1-pyrenyl)ethyl, PYE, aromatic 3-(p-nitrophenoxy)propyl, NPO, 3-phenoxypropyl, POP, 3-(p-methylmercaptophenoxy)propyl, CH3SPOP, 3-(pentafluorophenoxy)propyl, F5POP, and polymer-bonded silica/methyl methacrylate, MMA. C8, C18, POP, CH3SPOP, F5POP, and MMA were prepared in our laboratory, according to procedures described elsewhere.30,31 F13C9 was commercially available from Wako Pure Chemical Industries, Ltd. (http:// www.wako-chem.co.jp); PBB, PYE, and NPO were available from Nacalai Tesque (http://www.nacalai.co.jp/). Determination of the Retention Factors and Isotope Effects. The partition coefficient in chromatography is expressed as a retention factor:

k ) (tr - t0)/t0 where tr is the retention time of a solute on a given column and t0 is the retention time of a nonretained solute (“dead time”). The retention factor relates to free energy changes according to eq 1

ln K ) -∆G0/RT ) -∆H0/RT + ∆S0/R ) ln k - ln φ Figure 1. Structures of the stationary phases used.

all of the solutes we have chosen for study must interact with the stationary phase almost entirely, if not exclusively, by partitioning into the hydrophobic layer. Interaction with the silica particle core is believed to be minimal and most probably negligible for several reasons. First, the stationary phases are prepared with a high density of covalently attached organic chains that largely mask any residual, underivatized SiOH sites. Second, the isotopically substituted solutes studied are mainly nonpolar hydrocarbons. Third, our data show nearly identical values of the isotope effect per CH/CD bond for aliphatic alcohols as for aliphatic hydrocarbons (though differing among different stationary phases), indicating that the hydrocarbon parts of the chains are in closely similar environments regardless of the presence or absence of an attached alcohol group. Finally, our QSRR analysis shows that binding to all of these stationary phases is primarily if not entirely hydrophobic in nature. Thus the stationary phases we have studied have a hydrophobic solutebinding mechanism. Our isotope effect data do not of course provide information about the exact structure or ordering of the stationary phases and the bound solutes, but our data do tell us about the nature and extent of solute-stationary phase interactions. Experimental Section Equipment. The following analytical equipment was used for the chromatographic experiments: an LC-10AD pump at a flow rate 1 mL min-1 and SPD-10AV UV-Vis detector from Shimadzu (http:// www.shimadzu.com/) at λ ) 254 nm for the aromatic compounds and a JASCO 830-RI refractive index detector for aliphatic compounds (http://www.jasco.co.jp/). Constant column temperature was maintained with a water bath. Chromatograms (using mobile phases CH3OH/H2O, CH3CN/H2O) were collected and analyzed by a V-STATION chromatography data system (http://www.gls.co.jp/) on a PC. 13838 J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003

(1)

where K is the chromatographic binding equilibrium constant and φ is the column phase ratio, i.e., the ratio of the volumes of the stationary and the mobile phases. The retention factor of a methylene unit R(CH2), commonly used in chromatography, was also calculated. It is the ratio of the retention factors for two homologues, for example,

R(CH2) ) k1-pentanol-h11/k1-butanol-h9 and R(CD2) ) k1-pentanol-d11/k1-butanol-d9 (2) for the protiated and deuterated homologues, respectively. In addition, we measured R(CH2) using amylbenzene and butylbenzene. We have introduced in this work another unit that reflects the retention behavior of aromatic compounds

R(C4H2) ) knaphthalene-h8/kbenzene-h6

(3)

which we term the “retention factor of an aromatic unit”. The total isotope effect (TIE) is calculated to reflect the overall difference in chromatographic behavior of each isotopologue pair. It is free from any phase ratio influence (cf. eq 1, φ cancels out), since it is calculated as a ratio

TIE ) kH/kD where kH and kD are the retention factors for the protiated and deuterated isotopologue pair, respectively. Therefore, one may calculate direct free energy values for this process as well. The single isotope effect (%IE) reflects the average influence of a single H/D substitution and is given by

%IE ) 100[(kH/kD)1/n - 1]

(4)

where n is the number of D atoms substituted for H. (30) Kimata, K.; Hirose, T.; Moriuchi, K.; Hosoya, K.; Araki, T.; Tanaka, N. Anal. Chem. 1995, 67, 2556-2561. (31) Go, H.; Sudo, Y.; Hosoya, K.; Ikegami, T.; Tanaka, N. Anal. Chem. 1998, 70 (19), 4086-4093.

Isotope Effects on Hydrophobic Interactions

ARTICLES

Table 1. Dependence of Retention Capacity Factors, k, Isotope Effects TIEa (Total) and %IEb (Single), and Structural Unit Retention Factors R on Structure of Stationary Phasec compound

C8

C18

PYE

POP

NPO

F5POP

F13C9

CH3SPOP

PBB

MMA

benzene benzene-d6 TIE %IE toluene toluene-d8 TIE %IE naphthalene naphthalene-d8 TIE %IE anthracene anthracene-d10 TIE %IE nitrobenzene nitrobenzene-d5 TIE %IE cyclohexane cyclohexane-d12 TIE %IE hexane hexane-d14 TIE %IE octane octane-d18 TIE %IE 1-decanol 1-decanol-d21 TIE %IE 1-dodecanol 1-dodecanol-d25 TIE %IE R(CH2) R(C4H2) nD30d

0.872 0.847 1.030 0.486 1.358 1.310 1.037 0.450 1.768 1.696 1.042 0.522 3.727 3.532 1.055 0.539 0.629 0.617 1.019 0.385 3.481 3.373 1.032 0.263 5.055 4.901 1.031 0.221 11.887 11.387 1.044 0.239 5.278 5.009 1.054 0.249 12.036 11.312 1.064 0.248 1.510 2.028 1.404

1.507 1.455 1.036 0.586 2.609 2.498 1.044 0.545 3.847 3.644 1.056 0.680 11.595 10.762 1.077 0.748 0.887 0.865 1.025 0.503 7.836 7.519 1.042 0.345 12.136 11.565 1.049 0.345 35.406 33.264 1.064 0.347 10.292 9.588 1.073 0.338 29.850 27.400 1.089 0.343 1.680 2.553 1.441

0.850 0.825 1.030 0.499 1.342 1.286 1.044 0.534 2.745 2.621 1.047 0.579 11.629 10.895 1.067 0.654 2.016 1.941 1.039 0.760 1.986 1.901 1.045 0.365 2.608 2.468 1.057 0.395 6.320 5.878 1.075 0.404 4.307 3.963 1.087 0.397 9.573 8.680 1.103 0.393 1.540 3.229

0.695 0.679 1.024 0.390 0.977 0.949 1.030 0.364 1.624 1.575 1.031 0.384 3.978 3.829 1.039 0.382 0.835 0.818 1.021 0.413 1.599 1.553 1.030 0.243 1.856 1.795 1.034 0.239 3.487 3.334 1.046 0.250 2.166 2.062 1.050 0.234 3.991 3.764 1.060 0.234 1.380 2.337 1.513

0.514 0.505 1.018 0.294 0.733 0.716 1.024 0.293 1.470 1.435 1.024 0.302 4.890 4.737 1.032 0.318 0.950 0.936 1.015 0.298 1.010 0.988 1.022 0.184 1.166 1.136 1.026 0.186 2.119 2.044 1.037 0.200 1.349 1.297 1.040 0.187 2.390 2.274 1.051 0.199 1.350 2.860 1.576

0.801 0.791 1.013 0.209 1.250 1.233 1.014 0.171 2.045 2.003 1.021 0.260 5.168 5.020 1.029 0.291 0.952 0.946 1.006 0.126 1.695 1.689 1.004 0.030 2.510 2.502 1.003 0.023 5.063 5.022 1.008 0.045 2.508 2.485 1.009 0.044 4.907 4.862 1.009 0.037 1.380 2.553 1.427

0.420 0.417 1.007 0.120 0.564 0.562 1.004 0.045 0.451 0.445 1.013 0.168 0.511 0.502 1.018 0.178 0.379 0.379 1.000 0.000 1.348 1.369 0.985 -0.128 2.226 2.255 0.987 -0.093 3.968 4.061 0.977 -0.129 1.843 1.886 0.977 -0.110 3.248 3.345 0.971 -0.118 1.330 1.074 1.295e

0.980 0.956 1.025 0.414 1.387 1.343 1.033 0.404 2.581 2.493 1.035 0.435 8.087 7.724 1.047 0.460 1.292 1.263 1.023 0.456 1.937 1.869 1.036 0.298 2.224 2.130 1.044 0.309 4.492 4.235 1.061 0.328 2.680 2.517 1.065 0.299 5.266 4.882 1.079 0.303 1.410 2.634 1.573

1.393 1.343 1.037 0.611 2.426 2.316 1.047 0.582 7.299 6.909 1.056 0.688 46.854 43.472 1.078 0.752 1.897 1.841 1.030 0.601 3.101 2.955 1.049 0.403 3.840 3.619 1.061 0.425 10.014 9.275 1.080 0.427 5.580 5.118 1.090 0.413 13.930 12.563 1.109 0.414 1.590 5.240

0.704 0.698 1.008 0.143 0.890 0.881 1.011 0.127 1.809 1.796 1.007 0.090 4.172 4.145 1.007 0.065 1.032 1.026 1.006 0.116 0.951 0.939 1.012 0.106 0.971 0.956 1.015 0.111 1.630 1.599 1.019 0.107 0.693 0.672 1.032 0.147 1.167 1.125 1.038 0.147 1.296 2.570

a TIE ) k /k . b %IE ) 100[(k /k )1/n - 1]. c Cf. Figure 1; 70% methanol/water (vol/vol) mobile phase, 30 °C. d Experimentally measured refractive H D H D indices for the olefins used as the precursors for the stationary phases. e Measured at 20 °C.

Molecular Modeling and QSRR Calculations. Molecular modeling and structural descriptor calculations were done using HyperChem 5.1 Pro with ChemPlus 1.5 (http://www.hyper.com/). Protiated structures were first calculated by the extended Hu¨ckel method in order to estimate charges on the atoms and then optimized with the MM+ force field followed by RHF semiempirical (AM1) geometry optimization in a vacuum. The next step was placing the AM1 optimized structures inside a periodic box of water molecules and another optimization of geometry, where water was treated by a MM+ force field (classical approach) and the solute was treated by AM1 (semiempirical approach). For these optimized solute structures, calculations of the van der Waals surfaces and volumes were performed using the appropriate van der Waals radii values. Quantitative structure-chromatographic retention analyses by means of simple and multiple linear regression were performed on a PC machine using Statlets 1.1B (http://www.statlets.com/) and Prophet 5.0 (http://www.bbn.com/), freely available on the web for research purposes.

Results

This work investigates structural effects upon chromatographic binding equilibria, that is, equilibria for partitioning of solutes between aqueous and hydrophobic phases. By studying

the effects of changes in aqueous phase composition for different hydrophobic phases, of changes in hydrophobic phase for different aqueous phase compositions, and of different solutes as a function of both aqueous phase and hydrophobic phase composition, we are able to dissect important factors that contribute to the phenomenon of hydrophobic binding. To this end, we give below our results for (a) isotope effects upon binding, (b) linear free energy relationships, (c) structural unit retention factors R(CH2) for methylene and R(C4H2) for aromatics, and (d) temperature effects. Retention Factors. The retention data along with the total and single isotope values (TIE and %IE), for 10 isotopologue pairs on all stationary phases in 70% methanol/water mobile phase are given in Table 1. Results for the other mobile phase compositions are given in the supporting information (Table A). Samples injected were sufficiently small (ca. 10-20 µg aliphatic, ca. 1-2 µg aromatic) to approach infinite dilution in the mobile phase and also permit solubility in the aqueous mobile phases employed. The resulting chromatograms show minimal distortions in peak symmetry that might result from saturation of binding sites in the stationary phase. Representative chromatograms are shown in Figure 2. J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003 13839

ARTICLES

Turowski et al. Table 2. Results of QSRR Analysis (Eq 6) of 33 Test Compounds for a Series of Stationary Phasesa stationary phase SDb

C8 SD C18 SD PYE SD POP SD NPO SD F5POP SD F13C9 SD CH3SPOP SD PBB SD MMA SD

m

F

S

a

b

n

R2 c (p)

d -0.730 -0.364 -0.208 -1.304 1.393 0.987 0.060 0.036 0.032 0.056 0.059 (e10-4) -0.584 0.242 -0.679 -0.298 -1.509 1.632 0.991 0.066 0.077 0.066 0.036 0.079 0.071 (e10-4) d -1.218 0.498 -0.848 -1.154 1.428 0.963 0.107 0.065 0.057 0.099 0.105 (e10-4) d d -0.843 -0.371 -1.025 1.124 0.986 0.046 0.023 0.041 0.047 (e10-4) d -1.070 0.327 -0.330 -1.167 1.053 0.959 0.077 0.046 0.040 0.071 0.075 (e10-4) d -0.740 -0.139 -0.276 -1.130 1.218 0.954 0.088 0.067 0.049 0.078 0.093 (e10-4) d -0.677 -0.433 -0.438 -1.033 0.981 0.964 0.084 0.061 0.047 0.075 0.089 (e10-4) d -0.904 0.166 -0.419 -1.084 1.175 0.980 0.058 0.042 0.032 0.051 0.061 (e10-4) d -0.915 0.236 -0.476 -1.352 1.520 0.975 0.076 0.054 0.042 0.067 0.080 (e10-4) d -0.710 0.253 -0.143 -1.490 0.859 0.948 0.078 0.058 0.044 0.090 0.071 (e10-4)

Figure 2. Representative chromatograms of H/D separation in a 70% methanol/water mobile phase for seven isotopologue pairs on C18 and PYE phases. 1, Benzene-h6/d6; 2, toluene-h8/d8; 3, naphthalene-h8/d8; 4, anthracene-h10/d10; 5, cyclohexane-h12/d12; 6, n-octane-h18/d18; 7, dodecan-1ol-h25/d25.

a Cf. Figure 1; mobile phase 70% methanol/water, 30 °C. b SD ) standard deviation of the coefficient. c R2 ) determination coefficient of the multiple regression analysis; p ) significance level of the equation (eq 6). d Not included in correlation because significance level p was higher than 0.05.

The largest retention factors for aromatic solutes were exhibited by the aromatic PBB (pentabromobenzyloxypropyl) and PYE (pyrene) stationary phases. High k values are observed with the octadecyl phase (C18) for aliphatic compounds. Although the same mobile phase was used in each case, the capacity factors k still do not directly provide the actual binding equilibrium constants K and free energies, since every column differs according to its phase ratio φ (eq 1). Because φ is difficult to determine, accurate calculation of thermodynamic parameters remains problematic. Cancellation of O in Ratio-Based Structural Unit Retention Factors and Isotope Effects. Direct calculation of binding free energy becomes possible by defining structural unit retention factors R(CH2) for methylene and R(C4H2) for aromatics, which are computed from the retention data (eqs 2 and 3). Since (eq 1) K ) k/φ, the difficult-to-determine phase ratio φ cancels when ratios of k values are taken, with the result that the R factors equal true ratios of equilibrium constants K, as shown in eq 5. For the same reason, φ cancels out in isotope effects, so that the ratio of retention factors kH/kD equals KH/KD, the true equilibrium isotope effect on binding.

mechanisms of retention, such as hydrophobic binding,33 as well as in retention prediction.34 To provide a direct comparison of stationary phases, we carried out QSRR analysis for all 10 stationary phases. We employed the same series of 33 test solutes, under the same conditions, for each (see Table B in the supporting information) and used the structural descriptors proposed by Abraham35 to characterize the stationary phase properties.36 Abraham’s equation (eq 6), derived from the Kamlet-Taft solvatochromic parameters,37

k1/k2 ) K1φ/K2φ ) K1/K2 and kH/kD ) KHφ/KDφ ) KH/KD (5) For the same mobile phase composition, differences observed reflect the diverse properties and behavior of the stationary phases. In particular, we observed a similar order of elution of homologues upon all phases, while the H/D-pairs’ elution order was reversed in certain cases. Characterization of Stationary Phase Properties by Means of QSRR. Quantitative structure-chromatographic retention relationship (QSRR) analysis of retention data has an extrathermodynamic character32 and is useful in investigating molecular (32) Kaliszan, R. Structure and Retention in Chromatography. A Chemometric Approach; Harwood Academic Publishers: Amsterdam, 1997. 13840 J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003

log k ) m + νVx + Sπ2H + aΣR2H + bΣβ2H + FR2 (6) describes retention in terms of molecular properties of solutes and the chromatographic mobile and stationary phases, where m is a constant (intercept), Vx is the McGowan characteristic volume of the solute, R2H is the hydrogen-bond acidity of the solute, β2H is the hydrogen-bond basicity of the solute, π2H is the dipolarity-polarizability of the solute, and R2 is the excess molar refraction of the solute. Respective coefficients ν, S, a, b, and F reflect differences in properties between the stationary and mobile phases, to be discussed later. Table 2 presents the results of multiple regression analysis of retention data with 70% methanol/water mobile phase composition. The coefficients are related to the properties of both the stationary and mobile phases. However, if the same mobile phase is employed with different stationary phases, as in Table 2, the mobile phase properties cancel out in comparing the coefficients for different stationary phases. The results are presented for independent variables giving a significance level, p, e0.05. QSRR regression (33) Abraham, M. H.; Chadha, H. S.; Leitao, R. A. E.; Mitchell, R. C.; Lambert, W. J.; Kaliszan, R.; Nasal, A.; Haber, P. J. Chromatogr. 1997, 766, 3547. (34) (a) Tan, L. C.; Carr, P. W. J. Chromatogr. 1993, 656, 521-535. (b) Tan, L. C.; Carr, P. W.; Abraham, M. H. J. Chromatogr. 1996, 752, 1-18. (35) Abraham, M. H. Chem. Soc. ReV. 1993, 73-83. (36) Buszewski, B.; Gadzala-Kopciuch, R. M.; Markuszewski, M.; Kaliszan, R. Anal. Chem. 1997, 69, 3277-3284. (37) (a) Kamlet, M. J.; Taft, R. W. J. Am. Chem. Soc. 1976, 98, 377-383. (b) Taft, R. W.; Kamlet, M. J. J. Am. Chem. Soc. 1976, 98, 2886-2894.

Isotope Effects on Hydrophobic Interactions

ARTICLES

analysis was also performed including all variables regardless their significance levels (see Table C in the supporting information). Inclusion of all variables gives trends similar to those shown in Table 2. The coefficients indicate the sensitivity of solute binding to changes in stationary phase (while keeping the mobile phase constant at 70% methanol/water). For all stationary-phase materials,30 coefficients a (sensitivity to solute hydrogen bond acidity) and b (sensitivity to solute hydrogen bond basicity) are negative, while ν (sensitivity to solute McGowan characteristic volume) is always positive. Coefficient S (sensitivity to solute dipolarity-polarizability) is positive for PYE, NPO, and MMA but negative for C8 and C18 stationary phases, while F (sensitivity to solute excess molar refraction) is negative only for the fluorinated stationary phases. Table 1 includes the values of the available refractive indices of the precursors of the stationary phases. The refractive index is directly related to the molecular polarizability of compounds, according to the Lorentz-Lorentz formula (eq 7). The polarizability properties of the stationary phases may be qualitatively

R ) (nr2 - 1)/(nr2 + 2)

(7)

compared using these nD30 values to approximate the relative values for the chains of the stationary phase. Although the refractive indices of those expected to have the highest values, PBB and PYE, are not available, one notes that the precursors of the remaining aromatic phases have the greatest values, followed by the aliphatic C18 and C8. The precursor of the F13C9 phase has a refractive index smaller than that for water (nD30 ) 1.333). The negative coefficients a and b reflect the fact that polar interactions of solutes are much stronger in the aqueous mobile phase than in the stationary phase. The fact that a and b for all stationary phases are similar to that for the C18 phase indicates that solute binding by all phases is almost exclusively of a hydrophobic nature, especially involving London dispersion attractive interactions, as is already widely accepted for C18 stationary phases.36 That this similarity exists even for the NPO phase, which has a p-nitrophenyl group, indicates that the nitro groups are probably at the surface of this phase and interacting primarily with the aqueous mobile phase and that even in this case the binding of solutes is hydrophobic in nature. The data thus indicate that interactions, even of polar solutes, with the silica core of these stationary phases are most probably negligible. With our nonpolar solutes, if any interactions involving the silica core were present, they would most probably be with the polar methanol component of the aqueous phase and not with the bound solutes. Single Isotope Effect. As outlined in the Introduction, isotope effects upon binding constitute a unique probe of molecular interactions because isotopic substitution does not significantly affect potential energy. Potential energy effects cancel between isotopologues, meaning that isotope effects result essentially only from the effects of changes in interactions upon nuclear motions, especially vibrational frequencies. We have studied perdeuterated solutes as a means of obtaining RPLC separation of isotopologues and accurate values of isotope effects. For purposes of comparison and interpretation, isotope effects for a single deuterium substitution are needed, so we calculate averaged isotope effects per CH/CD bond from our directly observed total isotope effects, TIE, as %IE (eq 4). The

Figure 3. Logarithmic plot (ln-ln) of %IE [equivalent to ln(average isotope effect per H/D) × 100] vs R(CH2) (retention factor of a methylene unit, eq 2) for four representative solutes and the 10 stationary phases studied, mobile phase 70% methanol/water, 30 °C. Linear regressions are shown for the five aromatic stationary phases (see text). Stationary phases (cf. Figure 1) are MMA ) poly(methyl methacrylate), F13C9 ) 4,4-di(trifluoromethyl)5,5,6,6,7,7,7-heptafluoroheptyl, NPO ) 3-(p-nitrophenoxy)propyl, POP ) 3-phenoxypropyl, F5POP ) 3-(pentafluorophenoxy)propyl, CH3SPOP ) 3-(p-methylmercaptophenoxy)propyl, C8 ) octyl, PYE ) 2-(1-pyrenyl)ethyl, PBB ) 3-(pentabromobenzyloxy)propyl, and C18 ) octadecyl.

basis of this calculation is the usual approximation that, at ordinary temperatures, the isotope effect for each CH/CD bond contributes additively to the free energy (accurate if the isotope effect is controlled by zero-point energy differences) and thus multiplicatively to the TIE. That is, ln(TIE) ) ln(kH/kD) ) n ln(avg single CH/CD IE), where n is the number of CH/CD substitutions in the deuterated isotopologue, so that the average single CH/CD isotope effect ) (kH/kD)1/n, or, expressing this isotope effect on a percentage basis, %IE ) 100[(kH/kD)1/n 1]. In any molecule the average %IE cannot be exactly the same except for symmetry-related CH/CD bonds, so most of the molecules we have investigated involve finite differences among different types of bonds. However, since all are CH/CD bonds and are largely nonpolar, the interactions between them and the mobile as well as the stationary phase should be quite similar within a given molecule. We do see differences between aromatic and aliphatic CH/CD bond types (cf. Table 1), but the only molecule studied which contains both aromatic and aliphatic is toluene. For the other molecules at least, the use of averaged %IE should be a good approximation. This conclusion is strongly supported by the results, which show that %IE values are closely clustered for almost all compounds on all 10 stationary phases studied (see Figure 3 above). The fact that only relatively small differences in %IE are seen for different aliphatic chain lengths or for different aromatics demonstrates that there is little effect of structural variation within each series, aliphatic or aromatic. This is especially true for the aliphatic series and can be true only if the %IE for methyl protons and the different methylene protons in different chains are all nearly equal; if any type were significantly different, the average %IE should vary among the different compounds. The trends of isotope effects on binding relative to the binding affinity of each stationary phase for a single CH2 group are shown for representative solutes in Figure 3, plotted as %IE (eq 4) vs ln[R(CH2)] (eq 2). (Since logarithms are proportional J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003 13841

ARTICLES

Turowski et al.

to free energies, this is equivalent to a free energy vs free energy plot.) The retention factor of a methylene unit R(CH2) (eq 2) is taken as an approximate measure of the overall affinity or binding ability of each different stationary phase. A linear relationship is not necessarily expected but, as shown in Figure 3, the aromatic stationary phases alone give rather linear plots, suggesting the possibility that interactions with aromatic π electrons may play some role, while the nonaromatic stationary phases appear to form separate groups (see Discussion). Although isotope effects per CH/CD are small, amounting to fractions of a percent, their values have high reproducibility and thus differences in %IE values are highly significant. High reproducibilities are aided by the fact that isotopologue pairs may be injected simultaneously and separated in a single run, but the reproducibilities are high even in those cases where the isotope effect is small enough to necessitate separate determination of retention times by sequential injection of the protiated and deuterated isotopologues. Total isotope effects, TIE, for solutes containing several deuterium substitutions are typically several percent (cf. Table 1) and are reproducible to (0.1% in most cases. As a typical example, cyclohexane vs cyclohexaned12 on a C18 stationary phase gives a TIE ) 1.0422. A reproducibility of (0.001 gives, from eq 4, %IE ) 0.345 ( 0.008. Thus the differences in %IE observed for the different stationary phases are highly significant (range -0.128 to +0.403, with most differences well outside experimental reproducibility, cf. Table 1). It should also be emphasized that these seemingly small isotope effects nonetheless lead to complete or nearly complete chromatographic separation of isotopologues in most cases. These separations of course result from the very large numbers of theoretical plates characteristic of the HPLC columns employed. These large numbers of theoretical plates are the feature which allows us to determine small isotope effects with very high reproducibility and accuracy. Because of the unique nature of isotope effects, these isotope effects specifically measure only changes in the properties of the CH/CD bonds of the solute upon transfer from the aqueous mobile phase to the hydrophobic stationary phase. As a result, the %IE isolates separately the effect of the binding process upon just the solute, i.e., shows how the differences in binding interactions between the two phases affect the properties of the CH/CD bonds of the solute. The fact that the observed values of %IE are different from zero directly shows that the solute has different intermolecular interactions with the hydrophobic stationary phase than with the aqueous mobile phase. As a primary conclusion, our %IE data show that the solute does not interact with the stationary vs the mobile phase while remaining unperturbed; rather, the properties of the solute are altered, probably via changes in the vibrational frequencies of the CH/CD bonds, as a result of the intermolecular interactions of solute with stationary and/or mobile phases. Positive values of %IE show that the CH/CD bonds are less restricted in the stationary phase than in the mobile phasesusually caused by weakening of the bonds, in this case weaker in the stationary phase relative to the mobile phase. Importantly, such weakening in turn implies that the interactions of the solutes with the relatively nonpolar, hydrophobic stationary phase are stronger than those with the polar, aqueous mobile phase. This result, observed here for most solutes and a variety of stationary phases, 13842 J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003

Figure 4. Plot of %IE [equivalent to ln(average isotope effect per H/D) × 100] for benzene (open symbols) and log plot of R(CH2), the retention factor of a methylene unit (eq 2) (closed symbols) vs percent methanol in the methanol/water mobile phase for representative stationary phases, 30 °C. (Logarithmic plots are presented since logarithms are proportional to free energy differences, so that these plots are equivalent to plots of free energies vs percent methanol.)

is consistent with an active role of the hydrophobic phase in hydrophobic phenomena, i.e., with the idea that lipophilic phenomena make an important contribution to the hydrophobic effect. The values of %IE are higher for aromatic than for aliphatic CH/CD in all chromatographic systems examined. Moreover, for the aromatic isotopologues %IE increases with the molecular size, while it remains constant for the H/D alkanes. The largest isotope effects are found for the three stationary phases exhibiting the strongest binding as estimated by R(CH2), for aromatics the order of %IE being PBB > C18 > PYE, for aliphatics, PBB > PYE > C18. In contrast, the smallest isotope effects for aromatic solutes are found for the two stationary phases exhibiting the weakest binding as estimated by R(CH2), methyl methacrylate ester (MMA) and fluorinated F13C9. For aliphatic solutes, small isotope effects are found for MMA and F13C9, the latter even giving inverse isotope effects for aliphatic solutes, and fluorinated F5POP gives small isotope effects as well. These trends in %IE with different types of solutes and stationary phases are also consistent with an active role of the hydrophobic phase in hydrophobic phenomena. They suggest possible involvement of CH/CD---π (aromatic) interactions in the isotope effects and hence in the binding process (see Discussion). Effect of Mobile Phase Upon Retention and Isotope Effects. While it was not possible to examine most solutes over a wide range of solvent composition, it was possible to determine %IE for benzene over the range of water (0% methanol) up to 80% methanol/water, as shown in Figure 4 along with R(CH2), the retention factor of a methylene unit, on four representative stationary phases. There are significant changes in %IE which tend to parallel the trends in binding affinities for a methylene unit, with the exception that the %IE for the nonaromatic stationary phases C18 and F13C9 levels off at about 20% methanol and less. Entirely comparable effects are found with all solutes and stationary phases over the accessible range of 60-80% methanol for aromatic solutes and

Isotope Effects on Hydrophobic Interactions

∆G0,

ARTICLES

∆H0,

T∆S0

Figure 5. Gibbs free energy enthalpy and entropy term changes for transfer of aliphatic CH2 and aromatic C4H2 units from 70% methanol/water into four representative hydrophobic phases PYE, PBB, C18, and F13C9, 30 °C. Numerical values of ∆H0 are shown.

70-80% methanol for aliphatic solutes (complete data for all solutes are in Table A of supporting information). If the mobile phase simply provided an essentially noninteracting cage for the solute, the isotope effects would be independent of mobile phase composition. The fact that the isotope effects depend on mobile phase composition as well as on stationary phase demonstrates that the solute interacts with the mobile phase, too. Hence, these isotope effects show that the aqueous phase plays an active role as well, in addition to the important contribution to the hydrophobic effect from lipophilic phenomena (see the previous section). Temperature Effects on Binding and on Isotope Effects. Temperature effects were studied for all H/D isotopologues in 70% methanol/water mobile phase on four representative stationary phases. In all cases, the van’t Hoff plots for the measured chromatographic retention factor k were highly linear and show that the binding process is exothermic. As discussed above (see eq 5), k differs from the equilibrium constant for binding K as K ) k/φ. The phase ratio φ being difficult to determine, thermodynamic quantities involving k are not useful for comparison. However, φ cancels when ratios of k values are taken, meaning that the R factors and isotope effects equal true ratios of equilibrium constants K (cf. eq 5), so that meaningful thermodynamic quantities ∆G0, ∆H0, and ∆S0 can be calculated for the retention factor for a methylene unit R(CH2), eq 2, the retention factor for an aromatic unit R(C4H2), eq 3, and the isotope effect TIE. Figure 5 compares these thermodynamic quantities, taken from van’t Hoff plots on four stationary phases, for R(CH2) and R(C4H2); Figure 6, for the TIE of a representative arene, naphthalene, and alkane, octane. The fluoroalkane F13C9 stationary phase is unusual, possibly a reflection of the extremely low dispersive properties of this stationary phase. The free energy of binding is seen to be dominated by enthalpy rather than entropy in every case. ∆H0 is uniformly negative, indicating stronger interactions of the solute with the stationary phase than with the mobile phase, and ∆S0 is uniformly negative as well, indicating decreased freedom of motion of the solute in each stationary phase relative to the aqueous phase. These entropy results would not be expected if the primary source were freeing of solvent molecules organized around the solute in the aqueous phase. Instead, it appears that

Figure 6. Gibbs free energy ∆G0, enthalpy ∆H0, and entropy term T∆S0 components of the total isotope effect TIE for transfer of naphthalene and octane from 70% methanol/water into four representative hydrophobic phases, PYE, PBB, C18, and F13C9, 30 °C.

solute motional freedom is more restricted in the hydrophobic stationary phases. This restriction appears not to arise from solute conformational flexibility, since the (C4H2) unit is conformationally immobile and is contained within rigid aromatic systems. These results apply to (CH2) and (C4H2) units inserted into, respectively, alkane and aromatic structures, and so they do not necessarily parallel exactly the thermodynamic quantities for complete molecules. But they have the major feature of allowing comparison among different stationary phases of the inherent binding effects for structural units within molecules, helping to clarify the origins of the hydrophobic effect. The isotope effects are dominated by enthalpic contributions in all but one case, that of naphthalene binding to the unusual fluoroalkane F13C9 stationary phase. The latter case could represent an extreme in which interactions with the stationary phase are so weak that the entropic effect primarily reflects an aqueous phase (cavity) effect. Since the isotope effect is determined only by interactions of the solute, this result cannot be associated with any freeing of solvent molecules; rather, in this case an entropy-dominated isotope effect must reflect the contribution of changes in low-frequency vibrations or in rotations upon transfer from the aqueous to the hydrophobic phase. A plausible explanation is that, at least for the flat, rigid molecule naphthalene, solute rotations are inhibited in the aqueous solvent cage but less restricted in the hydrophobic phase. This hypothesis would imply that similar (small) aqueousphase effects would be present for all other stationary phases, too, but for all except the minimally interacting F13C9 phase, the aqueous-phase effect would be overwhelmed by effects associated with the stationary phase, and the isotope effect would be dominated by solute interactions with the hydrophobic phase. This interpretation is supported by the observation that the isotope effect for the flexible, nonrigid molecule octane is not dominated by entropy contributions, even with the F13C9 phase. The enthalpy-dominated nature of the remaining isotope effects is just as expected if their origin lies in changes in CH/ CD vibrations, especially the high-frequency stretching vibrations, since at the temperatures studied, changes in highfrequency vibrations give isotope effects that are dominated by changes in zero-point energy, an enthalpy term. The negative enthapies, corresponding to isotope effects >1.0, indicate a CH/ J. AM. CHEM. SOC.

9

VOL. 125, NO. 45, 2003 13843

ARTICLES

Turowski et al.

CD vibrational frequency decrease in both octane and naphthalene upon transfer from the aqueous phase into the hydrophobic phase. Differences in solvation would be expected to have small effects on vibrational frequencies, with the lower frequency arising from stronger interactions with solvent. In this case, decreased vibrational frequency in the hydrophobic phase implies stronger interactions of hydrophobic phases with CH/CD bonds than in the aqueous phase. Conversely, the isotope effect